Given the common use of rapid testing in GP practices, there was a need in data on performance at this particular "point of care" and to learn more about whether and how well these tests can be used in practice.
For the Roche SARS-CoV-2 Rapid Antigen Test we used in this study we could show a very high specificity, similar to the manufacturer specifications. The measured sensitivity of 78% was far lower than the manufacturer's reported sensitivity of 96.25% (21). The Paul Ehrlich Insitute (PEI - German Federal Institute for Vaccines and Biomedicines) determined an overall sensitivity of 45% only. The sensitivity for the validation samples with a Ct value ≤25 was still 88.9% (26). As shown by the PEI data, we were able to prove that the sensitivity decreased with increasing Ct-values, i.e. decreasing viral load. GPs and medical assistants found the POCT easy to implement into everyday practice. Even though its performance is lower than a PCR test, many GPs would use a rapid antigen test especially in situations with an immediate need for action.
Performance of the rapid antigen test
Professionally conducted antigen tests do not require manufacturer-independent validation to be approved for the German market (27). When investigated in independent clinical studies, the sensitivities of the Roche SARS-CoV-2 Rapid Antigen Test in symptomatic patients ranged from less than 63% to well over 85% (28–31). The different results can probably be explained by different study characteristics (test site, inclusion criteria, staff training, etc.). Therefore, external manufacturer-independent validation in different clinical settings is needed. This allows diagnostic accuracy and scope of a test result to be interpreted correctly.
Like other studies before, our results show that the test provides reliable results especially at very low Ct-values, i.e. at a high viral load (15). The viral load is dependent on the days since symptom onset. At the beginning of the infection, the viral load is particularly high and decreases over time (30, 32).The Ct-value alone is not sufficient to exclude risk of transmission. Infectivity is significantly dependent on SARS-CoV-2 gene copy number as well as genome integrity (33). However, the Ct-value correlates strongly with the viral load and for this reason, it can be a valuable tool for decision making and risk assessment (34). Above a Ct-value of >30, virus cultivation is difficult and infectivity seems unlikely (35). We showed that the POCT does not always provide reliable results even at a Ct-value of less than 30 (Fig. 3), a range of diagnostic inaccuracy prevails especially at Ct-values between 20 and 30.
Our results showed a sufficiently high specificity of antigen tests in primary care practices. However, even if positive test results can be interpreted with high probability as true positives, they should be confirmed by a RT-PCR. In contrast, the sensitivity of the test was only moderate and strongly depended on the viral load. Rapid antigen tests cannot provide information about the viral load, and thresholds to rule out contagiousness are difficult to define. Furthermore, the test cannot distinguish between a beginning and a declining infection. If a patient is at the beginning of the infection, the viral load may have increased after a short time and infection of other people cannot be ruled out. A negative test result cannot exclude a transmission-relevant infection with certainty and should not lead to a false sense of security (36).
The high mutation potential of the coronavirus is also a recurring topic of discussion. The mutations mostly affect the S-protein. Rapid tests, on the other hand, usually detect the N-protein. Thus, the tests should still be able to detect infection even in the presence of mutations (37). In the case of since recently dominating the omicron variant, however, initial data show that the sensitivity of the rapid test could be lower (38). Regular re-evaluations seems absolutely necessary to detect diagnostic deficiencies in time.
The PPV and NPV depend on the prevalence, and these parameters are highly dynamic as the pandemic is strongly fluctuating. This should still be kept in mind when interpreting the results.
Feasibility and benefits of the rapid antigen test
The additional test caused only minor logistical challenges for GPs and MA. Its regular implementation in everyday practice should be easily possible. GPs and MA considered a physician-performed test to be superior to lay testing because it is presumed to be more accurately administered. Vice versa, the public also seems to have the highest trust in physician-performed antigen tests. According to a survey in Germany, general practices are the preferred testing location for 33% of patients, pharmacies for 25% and testing centers for only 12% (14).
GPs considered positive test results helpful as they were perceived as very safe due to their high specificity and appropriate measures such as quarantine and specific medical care could be initiated immediately.
Strengths and weaknesses
Our RT-PCR samples were evaluated at different accredited laboratories. Despite of a consistently high quality standard, the Ct-values are only comparable to a limited extent. However, it can be assumed that the trends (high Ct-value=low viral load, low Ct-value=high viral load) are still correct. Furthermore, we could not consider the Ct-values of all positive patients because in some cases they were not documented by the practice or provided by the laboratories.
Two different nasopharyngeal swabs were taken consecutively from each patient. Differences in the two sample collections cannot be excluded.
MAs rated the feasibility of the tests worse than the GPs. In some cases, the additional documentation, which was needed for the study, was stated as a reason. In everyday practice without this documentation, the feasibility should achieve better ratings.
To further verify the results of the POCT and especially PCR testing, we offered antibody testing to all positive tested participants. Unfortunately, only one third of the patients took up this voluntary offer. However, although the performed serologies were able to confirm an infection in most of the cases, they also lead to the suggestion that the tests, including the RT-PCR test, do not always provide error-free results. Probably, a second additional PCR test could provide more reliability in future studies.